Ananya Iyer
Every project manager I speak to has now tried an AI assistant at least once, and most of them tried it the same way: they opened a chat window on a Monday, asked it to write the status report, read what came back, and quietly went back to writing the status report themselves.
That pattern is so consistent it stopped being interesting. What is interesting is what the people who stuck with it did differently, and it turns out to be four small decisions made before the first prompt is ever typed. None of them is technical.
Why the first attempt usually fails
The tool does the only thing it can: it writes something that looks like a status report. It has the right shape, the right headings and the right tone. It is also wrong in ways that take longer to find than writing the thing would have taken.
Pick one job, not the whole workflow
The job you want first has three properties: it happens often, it has a right answer you can check in under two minutes, and nobody will be hurt if it is wrong the first few times.
- Weekly or more, or it is not worth the setup
- The same shape each time, so one prompt keeps working
- Done by more than one person, so a good version spreads
Each of those has an obvious check. That check is the entire point: it is what turns a guess into a draft.
What to measure in the first month
Most teams count how often the tool was used, which goes up whether or not it is helping. Three numbers are worth having, and all three are about the check rather than the output.
Questions people ask
- Source to be added.
Published